XML Comes Of Age

By OLIVER RIST

Web content creators and application developers have dreamed up a slew of new applications for the browser paradigm--many of which were never meant to be supported by traditional HTML.

New apps have come in a variety of forms, such as Dynamic HTML and Cascading Style Sheets (CSS). But the entire HTML format suffers from several key limitations that can't be addressed by just another version change. For one thing, the ever-present standards war between Netscape and Microsoft has resulted in each company introducing its own proprietary extensions into HTML. This limits content design to only one browser unless designers jump through hoops to make it otherwise.

The resulting coding gymnastics also means that pushing the standards envelope is the status quo for Web developers. Pushing the World Wide Web Consortium (W3C) for a faster process in updating the HTML specification is futile, so designers are forced to publish their own tag extensions before they become approved.

Then there are the conceptual limitations to HTML. The name "hypertext" refers to exactly that: text. HTML was only intended to provide a way to easily publish basic text on the Web. But this hasn't extended well to other data types, like binaries, or just simple database connectivity. And it hasn't easily satisfied the needs of special industry vocabularies like mathematics, chemistry or finance. And, once you begin adding new data and symbols to HTML, the concept quickly outgrows traditional Web browsers. The way Web development is heading is to build Web applications in a platform independent language (i.e., Java) that can utilize Web protocols without resorting to a browser. Unfortunately, HTML has difficulty surviving outside a browser.

Along Comes XML

What makes the Extensible Markup Language so attractive is that it's compatible with HTML, yet manages to address the aforementioned problems. XML is a markup language meant to define data in a readable format without presentation constraints. In the same way that HTML is platform-independent, XML is database- and file format-independent. Plus, it's a truly extensible language, which means it doesn't necessarily need a browser.

XML allows you to develop apps in what is essentially a miniature markup language. XML ensures that anyone can read it, as long as they have a valid XML parser. To do this, developers need to construct what's called a document-type definition.

DTDs comprise the structure and rules of what you want to publish in XML. For example, if you want to define a DTD to publish movie scripts for a Hollywood studio, you could specify rules such as <characterX> to precede a line or to specify <scene_desc> to apply to scene descriptions.

These rules also can have values assigned to them regarding how they are nested or whether they need additional elements. Once your DTD is built, any interface that uses an XML parser can interpret your script document instance by simply referring to the rules in your DTD.

And if you're not interested in constructing your own DTD, you can easily use those created by others. A variety of them are available on the Web, such as the mathematical markup language (MML), synchronized multimedia integration language (SMIL), chemical markup language (CML) and even things such as Microsoft's Channel Definition Format (CDF).

Additionally, a DTD is not required with every instance of XML. For basic XML documents, any XML parser can infer the document rules simply by the way the elements are presented. DTDs are only required for more complicated document structures, but these are numerous in fields such as engineering, general science, law or medicine.

Will HTML Be Obsolete?

Whenever a new technology comes along, proponents of the old technology are always concerned that their baby is on the way out. Sometimes they're right, but usually the two just wind up co-existing. Such is the prognosis for HTML and XML.

While some have trumpeted that the arrival of XML sounds the death knell of HTML, that's not necessarily the case. Along with additional helpers, such as CSS, Java and the Extensible Style Sheet Language, the XML paradigm may eventually kill off HTML--but probably not.

HTML is still an easy way to present textual data visually--and significantly easier than by using XML or XSL. Most people using XML right now are happily using HTML in the same system. The idea is that XML is used to provide rich data-handling capabilities in a custom environment.

Our Players

The field of XML tools is a bit amorphous. As the new darling of the Web development world, XML is continuously being tugged in new directions. The result is that XML tool manufacturers have a difficult time keeping up with new trends. We tested the latest XML offerings from vendors who are attempting to keep up. SoftQuad sent us their XMetaL, which allows users with little or no knowledge of XML to create content--something that's critical to XML's evolution as a common data format.

The other products featured in this roundup are more traditional XML development tools that split the field between content creation and editing, and tools designed to build XML applications. These tools include Bluestone Software's Visual-XML, Extensibility's XML Authority, Icon Information-Systems' XML Spy and Vervet Logic's XML Pro. As you can see, the field is rife with new players and there is relatively little representation from the established players in HTML.

We evaluated these products on the basis of stable XML content creation, editing and development. No one tool was the best at all of these functions. Indeed, we expect to see future reviews focused on only one of these segments, not all three. As XML evolves, all these tools will undergo significant changes.

Visual-XML

Where SoftQuad's XMetaL does a polished job of rendering XML in a word processing environment, Bluestone Software Inc.'s Visual-XML has similar poise from an application development perspective. However, this is an important distinction for all of the competitors. Namely, that tools like XMetaL exist to create XML content and other tools, like Visual-XML, exist to build XML-based applications.

Bluestone includes a number of sample XML applications with Visual-XML, which illustrates just what an XML application is supposed to be. These applications are basically DTDs designed for specific functions rather than just document types. Bluestone includes a purchase order DTD that interfaces with a back-end database. Visual-XML's Help and Tutorial feature uses these applications to introduce programmers to the nuances of both XML and its development environment.

Visual-XML's workspace is definitely well-designed. The interface is equipped with everything you could possibly need to manipulate XML data and to combine it with outside data sources. The product has tools to perform full editing functions on raw XML code, DTD specifications, SQL queries and even Java.

Visual-XML's tools also integrate well with other utilities in Bluestone's stable, such as its popular Sapphire/Web application server platform, its XML-Server platform or its XwingML development tool, designed to merge Java and XML code.

Wizards make XML application building easier in Visual-XML as well. The Database Publishing wizard enables users to build XML documents from database sources. There's a Process wizard that lets users configure one-time or automatic publishing of XML documents and data into an XML server, such as Bluestone's XML-Server. Visual-XML is definitely optimized to take advantage of it.

Bluestone even places hooks to this product in its documentation, pointing to its XML-Server's Universal Listener Framework. ULF represents the communications portion of the document conversion engine described earlier.

An XML data source can be imported into the Server and ULF; this allows for publishing along any relevant communications protocol, including HTTP, SSL, FTP and even e-mail. There's also a scheduler so users can manage batch processing of XML documents along multiple destination lines.

The SQL editor is augmented by a database object browser that lets you get a full view of back-end database tables and columns, as well as data objects and static queries or stored procedures. We wouldn't want to do database administration from Visual-XML, but these tools are definitely enough to optimize an XML application for use with any SQL-compliant database. Bluestone has also done an admirable job with Mentor, its help system, which provides context-sensitive help as well as an interactive navigational aid that lets Mentor be constantly accessible as a toolbar.

For XML developers looking for a professional-level application development environment, Bluestone has definitely come through. Though it does pull a bit from the rest of Bluestone's XML product line, Visual-XML is still top-tier, both in features and price.

XML Authority

Web technologies are plagued by standards with the life span of fruit flies. This is even worse when a new technology is suddenly "discovered," the way that XML has been, and suddenly finds itself being tugged in directions it was never meant to travel.

An example might be with DTDs, which have been around XML and the Standard Generalized Markup Language (SGML) since the beginning, but are now considered unwieldy by many developers looking to exploit XML for more than just text documents. To aid these developers, the W3C just approved a draft specification of XML schemas in May 1999 (www.w3.org/TR/xmlschema-1/).

This has created problems for folks looking to build usable DTDs for various industries. If XML schemas render DTDs obsolete, the work already done will be in vain. But moving directly to schema creation also could prove futile if the standard doesn't catch on. Extensibility Inc. is one vendor looking to provide a solution to this dilemma with its XML Authority.

The standards for XML schemas are still evolving. It's a gray area, though, because DTDs also are the approved schema format. The difference is that a traditional XML DTD works just fine for document publishing but lacks good support for exchanging other types of data. So you can look at schemas as an extension of standard DTDs.

At this time, the W3C is considering four separate schema extensions, including XML-Data, the data definition markup language (DDML), document content definitions (DCDs), and schemas for object-oriented XML (SOX). We expect all these standards to be approved in time--but each will be geared toward different purposes and with an underlying foundation consisting of XML syntax.

The one you'll encounter first (aside from traditional DTDs) is XML-Data. Microsoft's Internet Explorer 5 (IE5) supports a subset of this spec--mainly, it has support for data-typing and name spaces as well as traditional XML syntax.

Where XML Authority steps in is that it's the first tool we've seen that is effectively standards-blind when it comes to DTDs. Basically, XML Authority regards each of these new schemas, as well as traditional DTDs, as essentially the same thing from a development point of view. While each standard has its own syntax and feature idiosyncrasies, there's still a wide base of commonality--and Extensibility's tool takes advantage of this in its workspace.

While we didn't find XML Authority's workspace to be as polished as Bluestone's, it's still open, fairly customizable and definitely aimed at XML developers rather than content editors. The workspace is divided into two basic windows.

The first window lets the developer view the entire DTD or schema in a hierarchical tree structure. In the bottom window, the work is divided into columns for easier editing. And, of course, once the work is done, you can save it either as a typical DTD or in any of the other formats mentioned earlier. Additionally, XML Authority can open and modify any existing schemas or DTDs and then save them, again in either native or converted format.

However, even though XML Authority supports all these standards in a visual format, you'll still need to be familiar with each DTD or schema's individual syntax and feature quirks. The package does offer a common set of tools as far as this paradigm can be extended, but once DTDs begin to diverge into individualities, the onus is on the developer to use these on his or her own, via hand coding.

Overall, XML Authority is an impressive first stab at creating a common base to a technology that is constantly shifting its focus. While we would have liked to see more specific tools for each available DTD or schema specification, it's understandable that Extensibility omitted these this time in favor of a future release.

As a viewer of DTDs and schemas, it's a great common viewing and basic editing package. For in-depth DTD and XML application design work, however, tools like Visual-XML and even XML Spy are still a step ahead in the features department.

XML Spy

Icon Information-Systems has released XML Spy version 2.5, which is a step ahead of XML Pro at a fraction of the cost. Available only as a Web download, XML Spy is quickly gaining attention in the shareware community.

XML Spy divides its workspace into three windows, similar to that of XML Pro. For instance, XML Spy's Grid View, which shows all the elements in a document in an hierarchical tree view, is similar to XML Pro's tree view. This view even allows in-place editing of any elements represented here.

The Source view lets you see the document in its XML source form and even augments that ability with user-customizable color highlighting of syntax elements. This is a great way to make quick-fix changes to a document by hand.

Finally, the Browser view allows you to render a document still under construction using IE5. What is really nice here is that this view fully supports CSS, XSL, Extensible HTML (XHTML) and XML namespaces, and can even be displayed inside the XML Spy workspace instead of toggling back and forth to IE5.

Finally, XML Spy's menus and toolbars offer the usual assortment of find/replace, cut and paste, and print editing tools. A really nice touch is the ability to save an XML document with invalid syntax or other problems. This might sound like a stretch, but when the five o'clock whistle blows, many an XML developer would rather save the document for tomorrow and go home than stay late figuring out every problem just so they can save their work.

Another key advantage to using XML Spy is that it can be used as a DTD editing and creation tool, as well as an XML document editor. XML Pro, by comparison, requires an extra-cost bundling option to do this. Even better, you can edit a DTD at the same time you're editing the XML document that uses it. This is a great cause-effect view, and XML Spy extends it to let you edit multiple XML documents simultaneously as well. XML Spy manages this via the Windows MDI and also makes good use of Windows Unicode support.

Unfortunately, this brings us to one weakness in XML Spy: It is Microsoft-bound. While we noticed the tool is faster than XML Pro (an easy determination considering it was written in C++, while Pro was written in Java), this also means it's available on only one platform, namely Windows. It also wasn't noticeably faster than Visual-XML.

Additionally, while we loved the fact that we could edit and validate DTDs, with XML Spy, we were a bit disturbed that it manages this feat only via Microsoft's MSXML Parser. For the moment, this is still a fairly standards-neutral solution, but, given Microsoft's track record in the open Web standards department, we're leery of MSXML as a long-term standard.

Another potential weakness is the product's lack of more robust creation tools. By this we mean features such as a database editing tool and publishing wizard, and support for CSS, XHTML or a Java editor. The reason this doesn't necessarily have to be a weakness, however, is because Icon doesn't position XML Spy as a content creation tool. XML Spy is meant as the perfect validation tool, both for your own content as well as that of others. And, at this, it excels.

Of course, while being Microsoft-bound has disadvantages, it also has an upside, such as XML Spy's ability to use RichEdit 3.0 in Windows 2000 both for faster display updates as well as foreign writing system support.

Overall, XML Spy represents one of the best values in a professional-level XML validation and editing package. While its lack of creation tools and close ties to Microsoft may become a disadvantage to some in the long run, this is by no means a certainty. For a powerful XML editing solution available right off the Web for a double digit price tag, this tool is hard to beat.

XMetaL

In this roundup, SoftQuad Software Inc.'s XMetaL stands out from the rest of the pack in that it's not just a generic XML development tool. This package was designed to let non-XML-proficient writers create XML content or data as easily as they would use a word processor. In fact, we'd classify the product as an XML word processor. While the results attainable by using it aren't perfect, they're pretty close. XMetaL represents a great step in the direction of broad audience XML usability.

The idea behind XMetaL is similar to the general idea behind XML: An XML tool for creating content that an XML conversion engine could distribute in a variety of formats, including HTML or as a database record. While a regular word processor could be used for this purpose as well, using something like XMetaL ensures error-free conversion by the central engine, since content is already in proper XML syntax.

This is where SoftQuad's expertise in SGML really pays off. Where its HTML development tools have often been criticized for their overly rigid adherence to specific HTML versions, this actually becomes an advantage in the XML world, since that technology has unforgiving syntax requirements as well.

Overall, this sounds like an obvious goal for XML editors, however. And it is. In fact, competition already exists from companies like ArborText and FrameMaker. But SoftQuad has stepped ahead of even these competitors in two important areas: Price and ease of use. XMetaL costs little more than many office suites. This lets an IT professional quickly build a usable DTD specification for the company's specific needs and then distribute the XMetaL package to all users with an inherent DTD already in place. All with a relatively small learning curve.

Similar to SoftQuad's various versions of HotMetaL, XMetaL looks much like the WYSIWYG word processing package it's designed to emulate. This basic word processing interface is augmented by four key additional tools. First, SoftQuad added great support for CSS by including a fully functioning editor. Then, developers also have access to a macro tool for using Document Object Model (DOM) and Component Object Model controls, which is useful for defining integration with databases and document management systems.

Finally, there is a Customization editor included for customizing keyboard strokes for specific DTDs as well as just general productivity.

For the content creation folks, XMetaL is an easy step into XML. Documents can be written in "normal" mode with XML syntax hidden from users or it can be displayed with XML coding "on." You also can view documents in a browser preview mode, but this feature is dependent exclusively on IE5.

Frankly, since the trend in XML is to display browser text in HTML, we're not sure how important a browser preview mode is anymore. The right side of every working screen displays all the relevant elements for the present text position and the Formatting bar has a drop down menu of all element names currently in use.

Basic word processing tools are all there, including a multilevel undo feature, cut and paste, and search and replace, as well as auto-save. In a nice show of maturity, SoftQuad made sure that its spellcheck feature checks only the written text, not the XML syntax codes. Overall, this is definitely the most easy to use XML workspace we've seen to date.

Our only complaint is that we would like to have seen an hierarchical view of the document in question, as well as its position within any larger document structure. This would have required an additional view on SoftQuad's part, but it definitely would have been worth it for administrators and developers alike.

For general-purpose development, XMetaL's addition of a fully W3C CSS2-compliant style sheet editor is an important feature. The CSS editor is basically there to handle basic visual presentation tasks for XML documents that will be distributed via a browser. However, like most XML presentation tools, this one is geared mostly towards text display, though its ability to do so with an eye towards the visually pleasing is definitely a step up from straight XML.

The editor allows XML writers to insert a style sheet link into their documents, which then refers the browser to that style sheet for visual processing. The CSS editor also is backed up by a Quick Styles feature, which lets developers easily create rules for font type and size, text alignment, and other text formatting features. Again, however, SoftQuad is heavily dependent on IE5 for these features, though the company plans to be browser independent in the future.

That's not to say the editor cannot create CSS-usable documents in other browsers--it is based on the W3C's CSS standard after all. But XMetaL's preview features will only display IE5 compatible documents for the moment, so additional Q&A testing is up to you.

In this roundup, XMetaL is in a class by itself. Should the evolution of straight digital text continue to evolve in XML's direction, XMetaL may well be the forebearer of all future word processing packages. As a present day product, however, XMetaL is an excellent combination of XML creation with CSS presentation facilities. While this relationship isn't perfect in version 1.0, it's certainly good enough for new users, and we expect it to improve markedly with time.

XML Pro

XML Pro 2.0's immediate predecessor was one of the first XML editors commercially available. Vervet Logic has built upon this pioneering effort to create a more sophisticated product, but again, this is one package definitely aimed at experienced developers, not newcomers.

XML Pro's interface is a clean visual development environment consisting of three main views. You'll see the usual toolbar and menu lists situated above a document window. This is divided into an hierarchical view that shows the entire document structure. Then on the right side is the Attribute View that shows all the information about a highlighted element. These change dynamically depending on the data point selected. Finally, there is an Edit window beneath the Attribute view, which lets editors add text content to the document.

While XML Pro's data manipulation features are similar to those in XMetaL, its interface isn't nearly as friendly. XML Pro does support drag and drop, undo, and cutting and pasting, but these are all functions aimed more at developers rather than writers. XML Pro basically assumes that data creation happens somewhere else and that the XML conversion process is where XML Pro steps in.

Something else you won't do with XML Pro is edit DTDs, though you can view them. Where other products build in DTD editors, XML Pro completely omits this feature, preferring instead to offer it as a bundle with Microstar's Near & Far Designer, which is a tool designed expressly for DTD editing. On the plus side, aside from viewing DTDs, XML Pro also parses them, which ensures that any imported DTDs are valid with regards to XML rules.

Unfortunately, if XML Pro finds a problem with an imported DTD, it simply won't open it--which may leave you in a lurch. XML files with errors, on the other hand, won't be ignored--unlike the approach taken by some other XML editing tools. Instead, XML Pro will open those files and immediately provide an error pointer marking the problem.

Where XML Pro shines is in its ability to combine Java and XML. The product includes IBM's XML4J parser, as well as support for Java Development Kit 1.2.1. Supporting XML4J means that the tool will be able to easily comply with the W3C's DOM as well as the SAX interface. Support for JDK 1.2.1 means XML Pro will have an easier time developing advanced XML/Java applications--something that is quickly becoming the darling of the Web development world.

Of course, support for JDK 1.2.1 also is a requirement since XML Pro is written in Java. But, on the upside, this also allows the product to run on Windows, Solaris and Linux. Making application development even easier, XML Pro also supports PCDATA and CDATA definitions, which are XML objects that contain binary data instead of just text.

One area where XML Pro could use more maturity, however, is in its help and tutorial features. Considering the rigid adherence of XML documents to a specific standard, a context-sensitive help system is necessary--not only in terms of learning the product, but in determining errors as well. While XML Pro's help system has improved over version 1.2, it's still a far cry from what we found in XMetaL or even Bluestone's Visual-XML.

Overall though, XML Pro is a full-featured XML editing package that just needs to decide whether it wants to be a content development tool or an application development product. While both these schools really depend on the XML specification, one product will not be able to serve both requirements.

At the moment, XML Pro seems aimed at the application development side, but with such a basic product in terms of features, it's almost a beginner's package. Picking sides and supporting that decision with more in-depth features and a real help system is what XML Pro needs to compete.

Oliver Rist is contributing technical editor at InternetWeek and technical director of Grand Central Networks Labs. He can be reached at orist@cmp.com.